At Bank of America, I built a LangGraph multi-agent system with supervisor, SQL, document retrieval, FAQ retrieval, and MCP-enabled search agents. Its five-stage ingestion pipeline achieved a 1.0000 Faithfulness score.
I also built a governed SQL MCP analytics platform with seven tools for controlled SQL Server access. It applies semantic governance across four business tables and 158 approved columns, with sensitive-column masking and schema-aware query generation.
At Bank of America, I engineered SQL safety gates for SELECT-only enforcement, injection detection, and semantic validation. In 18 benchmark cases, the system blocked all unsafe queries.
Previously, as an AI/ML Intern at 10xScale.ai, I developed an AI mock interviewer and a multi-agent recruitment chatbot. My projects include INT8 quantization benchmarks for DeiT and YOLOv3-Tiny, and a conversational resume editor using LangChain, Gemini, and PostgreSQL.
